Spatial reconstruction of Scottish summer temperatures from tree rings

Abstract

A detailed understanding of past temporal patterns and spatial expression of temperature variations is important to place recent anthropogenic climate change into a longer term context. In order to fill the current gap in our understanding of northwest European temperature variability, point-by-point principal component regression was used to reconstruct a spatial field of 0.5° temperature grids across Scotland. A sequence of reconstructions utilizing several combinations of detrending and disturbance correction procedures, and a selection of tree-ring parameters [including ring width (RW), maximum latewood density (MXD) and blue intensity (BI)] was used in an evaluation of reconstruction skill. The high resolution of the reconstructed field serves also as a diagnostic tool to spatially assess the temperature reconstruction potential of local chronologies. Best reconstruction results, reaching calibration r2 = 65.8% and verification r2 = 63.7% in central Scotland over the 1901–1976 period, were achieved using disturbance-corrected and signal-free detrended RW chronologies merged with BI data after low-pass (high-pass) filtering the RW (BI) chronologies. Calibration and verification r2 > 50% was attained for central, north and east Scotland, >40% in west and northwest, and >30% in southern Scotland with verification of nearly all grids showing some reconstruction skill. However, the full calibration potential of reconstructions outside central Scotland was reduced either due to residual disturbance trends undetected by the disturbance correction procedure or due to other climatic or non-climatic factors which may have adversely affected the strength of the climate signal.